nvme: sanitize KATO handling
v2 cover nvme: sanitize KATO handling
Hi all,
one of our customer had been running into a deadlock trying to terminate outstanding KATO commands during reset.
Looking closer at it, I found that we never actually _track_ if a KATO command is submitted, so we might happily be sending several KATO commands to the same controller simultaneously.
Also, I found it slightly odd that we signal a different KATO value to the controller than what we're using internally; I would have thought that both sides should agree on the same KATO value. And even that wouldn't be so bad, but we really should be using the KATO value we annouonced to the controller when setting the request timeout.
With these patches I attempt to resolve the situation; the first patch ensures that only one KATO command to a given controller is outstanding. With that the delay between sending KATO commands and the KATO timeout are decoupled, and we can follow the recommendation from the base spec to send the KATO commands at half the KATO timeout intervals. And to aid userspace detecting persistent controllers we should be displaying the KATO value in sysfs, too.
As usual, comments and reviews are welcome.
Changes to v1:
- Include reviews from Christoph
- Include reviews from Chao Leng
- Add new patch to display KATO value in sysfs
Hannes Reinecke (3):
nvme: fixup kato deadlock
nvme: sanitize KATO setting
nvme: add 'kato' sysfs attribute
drivers/nvme/host/core.c | 21 ++++++++++++++++----- drivers/nvme/host/fabrics.c | 2 +- drivers/nvme/host/nvme.h | 10 +++++++++- 3 files changed, 26 insertions(+), 7 deletions(-)
Signature · 7 lines
2.29.2 _______________________________________________ Linux-nvme mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-nvme
PATCH v2 1/3 nvme: fixup kato deadlock
A customer of ours has run into this deadlock with RDMA:
- The ka_work workqueue item is executed
- A new ka_work workqueue item is scheduled just after that.
- Now both, the kato request timeout _and_ the workqueue delay will execute at roughly the same time
- If the timing is correct the workqueue executes _before_ the kato request timeout triggers
- Kato request timeout triggers, and starts error recovery
- error recovery deadlocks, as it needs to flush the kato workqueue item; this is stuck in nvme_alloc_request() as all reserved tags are in use.
The reserved tags would have been freed up later when cancelling all outstanding requests in the queue:
nvme_stop_keep_alive(&ctrl->ctrl); nvme_rdma_teardown_io_queues(ctrl, false); nvme_start_queues(&ctrl->ctrl); nvme_rdma_teardown_admin_queue(ctrl, false); blk_mq_unquiesce_queue(ctrl->ctrl.admin_q);
but as we're stuck in nvme_stop_keep_alive() we'll never get this far.
To fix this a new controller flag 'NVME_CTRL_KATO_RUNNING' is added which will short-circuit the nvme_keep_alive() function if one keep-alive command is already running.
Additionally we should be allocating the KATO request with BLK_MQ_REQ_NOWAIT as we must not block on request allocation; if we cannot get a request we cannot determine if the connection is healthy, and need to reset it anyway.
--- drivers/nvme/host/core.c | 10 ++++++++-- drivers/nvme/host/nvme.h | 1 + 2 files changed, 9 insertions(+), 2 deletions(-)
drivers/nvme/host/core.c +8 -2
@@ -1207,6 +1207,7 @@ static void nvme_keep_alive_end_io(struct request *rq, blk_status_t status)bool startka = false;blk_mq_free_request (rq); +clear_bit (NVME_CTRL_KATO_RUNNING , &ctrl->flags);if (status) { dev_err(ctrl->device,
@@ -1229,10 +1230,15 @@ static int nvme_keep_alive(struct nvme_ctrl *ctrl) {struct request *rq; +if (test_and_set_bit (NVME_CTRL_KATO_RUNNING , &ctrl->flags)) +return 0 ; + rq =nvme_alloc_request (ctrl->admin_q, &ctrl->ka_cmd, -BLK_MQ_REQ_RESERVED ); -if (IS_ERR (rq)) +BLK_MQ_REQ_RESERVED |BLK_MQ_REQ_NOWAIT ); +if (IS_ERR (rq)) { +clear_bit (NVME_CTRL_KATO_RUNNING , &ctrl->flags);return PTR_ERR (rq); + } rq->timeout = ctrl->kato *HZ ; rq->end_io_data = ctrl;
drivers/nvme/host/nvme.h +1 -0
@@ -344,6 +344,7 @@ struct nvme_ctrl {int nr_reconnects;unsigned long flags;#define NVME_CTRL_FAILFAST_EXPIRED 0 +#define NVME_CTRL_KATO_RUNNING 1struct nvmf_ctrl_options *opts;struct page *discard_page;
Signature · 7 lines
2.29.2 _______________________________________________ Linux-nvme mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-nvme
Christoph Hellwig
maintainerRe: [PATCH 1/3] nvme: fixup kato deadlock
drivers/nvme/host/core.c static int nvme_keep_alive(struct nvme_ctrl *ctrl) · line 1241
Quoted 11 lines · Hannes Reinecke
drivers/nvme/host/core.c +7
-2+if (test_and_set_bit (NVME_CTRL_KATO_RUNNING , &ctrl->flags)) +return 0 ; + rq =nvme_alloc_request (ctrl->admin_q, &ctrl->ka_cmd, -BLK_MQ_REQ_RESERVED ); -if (IS_ERR (rq)) +BLK_MQ_REQ_RESERVED |BLK_MQ_REQ_NOWAIT ); +if (IS_ERR (rq)) { +clear_bit (NVME_CTRL_KATO_RUNNING , &ctrl->flags);return PTR_ERR (rq); + }
Adding BLK_MQ_REQ_NOWAIT should be a separate prep patch, together with reducing the number of reserved tags.
Also why do we still need the extra test_and_set_bit with the NOWAIT allocation?
_______________________________________________
Linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme
Re: [PATCH 1/3] nvme: fixup kato deadlock
Quoted 16 lines · Christoph Hellwig
drivers/nvme/host/core.c static int nvme_keep_alive(struct nvme_ctrl *ctrl) · line 1241
Quoted 11 lines · Hannes Reinecke
drivers/nvme/host/core.c +7
-2+if (test_and_set_bit (NVME_CTRL_KATO_RUNNING , &ctrl->flags)) +return 0 ; + rq =nvme_alloc_request (ctrl->admin_q, &ctrl->ka_cmd, -BLK_MQ_REQ_RESERVED ); -if (IS_ERR (rq)) +BLK_MQ_REQ_RESERVED |BLK_MQ_REQ_NOWAIT ); +if (IS_ERR (rq)) { +clear_bit (NVME_CTRL_KATO_RUNNING , &ctrl->flags);return PTR_ERR (rq); + }Adding
BLK_MQ_REQ_NOWAITshould be a separate prep patch, together with reducing the number of reserved tags.
Okay.
But why would we need to reduce the number of tags? It's not that we're changing anything with the allocation, we're just guaranteed to fail if for some reason the stack messes up.
Quoted 3 lines
Also why do we still need the extra
test_and_set_bitwith the NOWAIT allocation?
This is essentially a safeguard; there is nothing in the current code telling us if a KATO command is in flight or not.
And I really do hate doing pointless work (like allocating) commands if we don't need to.
Plus it'll differentiate between legit callers of nvme_start_keep_alive() (which can be called at any time, and hence might be adding the ka_work element just after the previous one had submitted the KATO command), and real failures like failing to allocate the KATO command itself, which points to a real issue in the stack as normally there should be enough reserved commands.
But yeah, I'll split if off in two patches.
Cheers,
Hannes
Signature · 9 lines
Dr. Hannes Reinecke Kernel Storage Architect [email protected] +49 911 74053 688 SUSE Software Solutions Germany GmbH, Maxfeldstr. 5, 90409 Nürnberg HRB 36809 (AG Nürnberg), GF: Felix Imendörffer _______________________________________________ Linux-nvme mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-nvme
Christoph Hellwig
maintainerRe: [PATCH 1/3] nvme: fixup kato deadlock
Quoted 7 lines · Hannes Reinecke
Quoted 3 lines
Adding
BLK_MQ_REQ_NOWAITshould be a separate prep patch, together with reducing the number of reserved tags.Okay.
But why would we need to reduce the number of tags? It's not that we're changing anything with the allocation, we're just guaranteed to fail if for some reason the stack messes up.
Because we could otherwise still have to keep alive requests, or one keep a live and one connect requests allocated at the same time.
Quoted 9 lines
Quoted 4 lines
Also why do we still need the extra
test_and_set_bitwith the NOWAIT allocation?This is essentially a safeguard; there is nothing in the current code telling us if a KATO command is in flight or not.
And I really do hate doing pointless work (like allocating) commands if we don't need to.
Quoted 6 lines
Plus it'll differentiate between legit callers of
nvme_start_keep_alive()(which can be called at any time, and hence might be adding theka_workelement just after the previous one had submitted the KATO command), and real failures like failing to allocate the KATO command itself, which points to a real issue in the stack as normally there should be enough reserved commands.
You can distinguish them by the error code from blk_mq_alloc_request.
_______________________________________________
Linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme
Re: [PATCH 1/3] nvme: fixup kato deadlock
Quoted 4 lines · Christoph Hellwig
Quoted 3 lines · Hannes Reinecke
Quoted 3 lines
Adding
BLK_MQ_REQ_NOWAITshould be a separate prep patch, together with reducing the number of reserved tags.
Quoted 8 lines
Quoted 4 lines
Okay.
But why would we need to reduce the number of tags? It's not that we're changing anything with the allocation, we're just guaranteed to fail if for some reason the stack messes up.Because we could otherwise still have to keep alive requests, or one keep a live and one connect requests allocated at the same time.
But that's what we have today.
So you're saying that we should have only _1_ reserved request, ie either connect _or_ keep-alive?
Sure we can do that; I had been a bit hesitant to do that seeing that it had been that way since the beginning.
But if you say ...
Quoted 9 lines
Quoted 9 lines
Quoted 4 lines
Also why do we still need the extra
test_and_set_bitwith the NOWAIT allocation?This is essentially a safeguard; there is nothing in the current code telling us if a KATO command is in flight or not.
And I really do hate doing pointless work (like allocating) commands if we don't need to.
Quoted 9 lines
Quoted 6 lines
Plus it'll differentiate between legit callers of
nvme_start_keep_alive()(which can be called at any time, and hence might be adding theka_workelement just after the previous one had submitted the KATO command), and real failures like failing to allocate the KATO command itself, which points to a real issue in the stack as normally there should be enough reserved commands.You can distinguish them by the error code from
blk_mq_alloc_request.
True. Will be dropping that patch, then.
Cheers,
Hannes
Signature · 9 lines
Dr. Hannes Reinecke Kernel Storage Architect [email protected] +49 911 74053 688 SUSE Software Solutions Germany GmbH, Maxfeldstr. 5, 90409 Nürnberg HRB 36809 (AG Nürnberg), GF: Felix Imendörffer _______________________________________________ Linux-nvme mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-nvme
Re: [PATCH 1/3] nvme: fixup kato deadlock
Quoted 13 lines · Hannes Reinecke
Quoted 4 lines · Christoph Hellwig
Quoted 3 lines · Hannes Reinecke
Quoted 3 lines
Adding
BLK_MQ_REQ_NOWAITshould be a separate prep patch, together with reducing the number of reserved tags.Quoted 8 lines
Quoted 4 lines
Okay.
But why would we need to reduce the number of tags? It's not that we're changing anything with the allocation, we're just guaranteed to fail if for some reason the stack messes up.Because we could otherwise still have to keep alive requests, or one keep a live and one connect requests allocated at the same time.
Quoted 8 lines
But that's what we have today.
So you're saying that we should have only _1_ reserved request, ie either connect _or_ keep-alive?
Sure we can do that; I had been a bit hesitant to do that seeing that it had been that way since the beginning.But if you say ...
Actually, having thought a bit more, I'm not sure if we should change that. Changing the number of reserved commands _will_ have an impact on error recovery, up to the point that we cannot issue a 'connect' command if a keep-alive command is still running.
Which means we might be seeing spurious connection failures during recovery, something which I'd be very cautious about.
Cheers,
Hannes
Signature · 9 lines
Dr. Hannes Reinecke Kernel Storage Architect [email protected] +49 911 74053 688 SUSE Software Solutions Germany GmbH, Maxfeldstr. 5, 90409 Nürnberg HRB 36809 (AG Nürnberg), GF: Felix Imendörffer _______________________________________________ Linux-nvme mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-nvme
Christoph Hellwig
maintainerRe: [PATCH 1/3] nvme: fixup kato deadlock
Quoted 6 lines · Hannes Reinecke
Actually, having thought a bit more, I'm not sure if we should change that. Changing the number of reserved commands _will_ have an impact on error recovery, up to the point that we cannot issue a 'connect' command if a keep-alive command is still running.
Which means we might be seeing spurious connection failures during recovery, something which I'd be very cautious about.
We always do a blk_mq_tagset_busy_iter that cancels all requests before reconnecting.
_______________________________________________
Linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme
Re: [PATCH 1/3] nvme: fixup kato deadlock
Quoted 11 lines · Christoph Hellwig
Quoted 6 lines · Hannes Reinecke
Actually, having thought a bit more, I'm not sure if we should change that. Changing the number of reserved commands _will_ have an impact on error recovery, up to the point that we cannot issue a 'connect' command if a keep-alive command is still running.
Which means we might be seeing spurious connection failures during recovery, something which I'd be very cautious about.We always do a
blk_mq_tagset_busy_iterthat cancels all requests before reconnecting.
Right; and as we're now failing allocations we should be good. Let's see whether it holds up during testing.
Cheers,
Hannes
Signature · 9 lines
Dr. Hannes Reinecke Kernel Storage Architect [email protected] +49 911 74053 688 SUSE Software Solutions GmbH, Maxfeldstr. 5, 90409 Nürnberg HRB 36809 (AG Nürnberg), Geschäftsführer: Felix Imendörffer _______________________________________________ Linux-nvme mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-nvme
PATCH v2 2/3 nvme: sanitize KATO setting
In reply to Hannes ReineckeAccording to the NVMe base spec the KATO commands should be sent at half of the KATO interval, to properly account for round-trip times.
As we now will only ever send one KATO command per connection we can easily use the recommended values.
This also fixes a potential issue where the request timeout for the KATO command does not match the value in the connect command, which might be causing spurious connection drops from the target.
--- drivers/nvme/host/core.c | 9 ++++++--- drivers/nvme/host/fabrics.c | 2 +- drivers/nvme/host/nvme.h | 9 ++++++++- 3 files changed, 15 insertions(+), 5 deletions(-)
drivers/nvme/host/core.c +6 -3
@@ -1223,7 +1223,8 @@ static void nvme_keep_alive_end_io(struct request *rq, blk_status_t status) startka = true;spin_unlock_irqrestore (&ctrl->lock, flags);if (startka) -queue_delayed_work (nvme_wq, &ctrl->ka_work, ctrl->kato *HZ ); +queue_delayed_work (nvme_wq, &ctrl->ka_work, +NVME_KATO_DELAY (ctrl->kato) *HZ ); }static int nvme_keep_alive (struct nvme_ctrl *ctrl)
@@ -1258,7 +1259,8 @@ static void nvme_keep_alive_work(struct work_struct *work)dev_dbg (ctrl->device,"reschedule traffic based keep-alive timer\n" ); ctrl->comp_seen = false; -queue_delayed_work (nvme_wq, &ctrl->ka_work, ctrl->kato *HZ ); +queue_delayed_work (nvme_wq, &ctrl->ka_work, +NVME_KATO_DELAY (ctrl->kato) *HZ );return ; }
@@ -1275,7 +1277,8 @@ static void nvme_start_keep_alive(struct nvme_ctrl *ctrl)if (unlikely (ctrl->kato ==0 ))return ; -queue_delayed_work (nvme_wq, &ctrl->ka_work, ctrl->kato *HZ ); +queue_delayed_work (nvme_wq, &ctrl->ka_work, +NVME_KATO_DELAY (ctrl->kato) *HZ ); }void nvme_stop_keep_alive (struct nvme_ctrl *ctrl)
drivers/nvme/host/fabrics.c +1 -1
@@ -382,7 +382,7 @@ int nvmf_connect_admin_queue(struct nvme_ctrl *ctrl)* and add a grace period for controller kato enforcement */ cmd.connect.kato = ctrl->kato ? -cpu_to_le32 ((ctrl->kato +NVME_KATO_GRACE ) *1000 ) :0 ; +cpu_to_le32 (ctrl->kato *1000 ) :0 ;if (ctrl->opts->disable_sqflow) cmd.connect.cattr |=NVME_CONNECT_DISABLE_SQFLOW ;
drivers/nvme/host/nvme.h +8 -1
@@ -27,7 +27,14 @@ extern unsigned int admin_timeout;#define NVME_ADMIN_TIMEOUT (admin_timeout * HZ)#define NVME_DEFAULT_KATO 5 -#define NVME_KATO_GRACE 10 + +/* +* The recommended frequency for KATO commands +* according to NVMe 1.4 section 7.12.1: +* The host should send Keep Alive commands at half of the +* Keep Alive Timeout accounting for transport roundtrip times [..]. +*/ +#define NVME_KATO_DELAY (k) ((k) >> 1)#ifdef CONFIG_ARCH_NO_SG_CHAIN #define NVME_INLINE_SG_CNT 0
Signature · 7 lines
2.29.2 _______________________________________________ Linux-nvme mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-nvme
Re: [PATCH 2/3] nvme: sanitize KATO setting
Quoted 21 lines · Hannes Reinecke
According to the NVMe base spec the KATO commands should be sent at half of the KATO interval, to properly account for round-trip times.
As we now will only ever send one KATO command per connection we can easily use the recommended values.
This also fixes a potential issue where the request timeout for the KATO command does not match the value in the connect command, which might be causing spurious connection drops from the target.Signed-off-by Hannes Reinecke <[email protected]>--- drivers/nvme/host/core.c | 9 ++++++--- drivers/nvme/host/fabrics.c | 2 +- drivers/nvme/host/nvme.h | 9 ++++++++- 3 files changed, 15 insertions(+), 5 deletions(-)drivers/nvme/host/core.c +0
-0
drivers/nvme/host/core.c static void nvme_keep_alive_end_io(struct request *rq, blk_status_t status) · line 1227
Quoted 6 lines
drivers/nvme/host/core.c +2
-1@@ -1223,7 +1223,8 @@ static void nvme_keep_alive_end_io(struct request *rq, blk_status_t status) startka = true;spin_unlock_irqrestore (&ctrl->lock, flags);if (startka) -queue_delayed_work (nvme_wq, &ctrl->ka_work, ctrl->kato *HZ ); +queue_delayed_work (nvme_wq, &ctrl->ka_work, +NVME_KATO_DELAY (ctrl->kato) *HZ );
* HZ and then >> 1 is a better choice. do like this: NVME_KATO_DELAY(ctrl->kato * HZ).
Quoted 47 lines
drivers/nvme/host/core.c +4
-2}static int nvme_keep_alive (struct nvme_ctrl *ctrl)@@ -1258,7 +1259,8 @@ static void nvme_keep_alive_work(struct work_struct *work)dev_dbg (ctrl->device,"reschedule traffic based keep-alive timer\n" ); ctrl->comp_seen = false; -queue_delayed_work (nvme_wq, &ctrl->ka_work, ctrl->kato *HZ ); +queue_delayed_work (nvme_wq, &ctrl->ka_work, +NVME_KATO_DELAY (ctrl->kato) *HZ );return ; }@@ -1275,7 +1277,8 @@ static void nvme_start_keep_alive(struct nvme_ctrl *ctrl)if (unlikely (ctrl->kato ==0 ))return ; -queue_delayed_work (nvme_wq, &ctrl->ka_work, ctrl->kato *HZ ); +queue_delayed_work (nvme_wq, &ctrl->ka_work, +NVME_KATO_DELAY (ctrl->kato) *HZ ); }void nvme_stop_keep_alive (struct nvme_ctrl *ctrl)drivers/nvme/host/fabrics.c +1
-1@@ -382,7 +382,7 @@ int nvmf_connect_admin_queue(struct nvme_ctrl *ctrl)* and add a grace period for controller kato enforcement */ cmd.connect.kato = ctrl->kato ? -cpu_to_le32 ((ctrl->kato +NVME_KATO_GRACE ) *1000 ) :0 ; +cpu_to_le32 (ctrl->kato *1000 ) :0 ;if (ctrl->opts->disable_sqflow) cmd.connect.cattr |=NVME_CONNECT_DISABLE_SQFLOW ;drivers/nvme/host/nvme.h +1
-1@@ -27,7 +27,14 @@ extern unsigned int admin_timeout;#define NVME_ADMIN_TIMEOUT (admin_timeout * HZ)#define NVME_DEFAULT_KATO 5 -#define NVME_KATO_GRACE 10 +
drivers/nvme/host/nvme.h extern unsigned int admin_timeout; · line 40
Quoted 10 lines
drivers/nvme/host/nvme.h +7
-0+/* +* The recommended frequency for KATO commands +* according to NVMe 1.4 section 7.12.1: +* The host should send Keep Alive commands at half of the +* Keep Alive Timeout accounting for transport roundtrip times [..]. +*/ +#define NVME_KATO_DELAY (k) ((k) >> 1)#ifdef CONFIG_ARCH_NO_SG_CHAIN #define NVME_INLINE_SG_CNT 0
_______________________________________________
Linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme
Christoph Hellwig
maintainerRe: [PATCH 2/3] nvme: sanitize KATO setting
Quoted 19 lines · Chao Leng
Quoted 16 lines · Hannes Reinecke
According to the NVMe base spec the KATO commands should be sent at half of the KATO interval, to properly account for round-trip times.
As we now will only ever send one KATO command per connection we can easily use the recommended values.
This also fixes a potential issue where the request timeout for the KATO command does not match the value in the connect command, which might be causing spurious connection drops from the target.Signed-off-by Hannes Reinecke <[email protected]>--- drivers/nvme/host/core.c | 9 ++++++--- drivers/nvme/host/fabrics.c | 2 +- drivers/nvme/host/nvme.h | 9 ++++++++- 3 files changed, 15 insertions(+), 5 deletions(-)
Quoted 12 lines
drivers/nvme/host/core.c static void nvme_keep_alive_end_io(struct request *rq, blk_status_t status) · line 1227
Quoted 11 lines
drivers/nvme/host/core.c +2
-1@@ -1223,7 +1223,8 @@ static void nvme_keep_alive_end_io(struct request *rq, blk_status_t status) startka = true;spin_unlock_irqrestore (&ctrl->lock, flags);if (startka) -queue_delayed_work (nvme_wq, &ctrl->ka_work, ctrl->kato *HZ ); +queue_delayed_work (nvme_wq, &ctrl->ka_work, +NVME_KATO_DELAY (ctrl->kato) *HZ );* HZ and then >> 1 is a better choice. do like this:
NVME_KATO_DELAY(ctrl->kato * HZ).
Also there is no reason to not include the "* HZ" in NVME_KATO_DELAY, as all callers do it anyway. Also please turn NVME_KATO_DELAY into a normal inline function using normal lower case naming. Alternatively we could also just do a helper for the whole queue_delayed_work, which might actually end up even cleaner.
_______________________________________________
Linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme
Sagi Grimberg
maintainerRe: [PATCH 2/3] nvme: sanitize KATO setting
In reply to Hannes Reineckedrivers/nvme/host/fabrics.c int nvmf_connect_admin_queue(struct nvme_ctrl *ctrl) · line 385
Quoted 11 lines
drivers/nvme/host/fabrics.c +1
-1@@ -382,7 +382,7 @@ int nvmf_connect_admin_queue(struct nvme_ctrl *ctrl)* and add a grace period for controller kato enforcement */ cmd.connect.kato = ctrl->kato ? -cpu_to_le32 ((ctrl->kato +NVME_KATO_GRACE ) *1000 ) :0 ; +cpu_to_le32 (ctrl->kato *1000 ) :0 ;
There is no need for the conditional anymore..
_______________________________________________
Linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme
Re: [PATCH 2/3] nvme: sanitize KATO setting
In reply to Hannes ReineckeQuoted 70 lines · Hannes Reinecke
According to the NVMe base spec the KATO commands should be sent at half of the KATO interval, to properly account for round-trip times.
As we now will only ever send one KATO command per connection we can easily use the recommended values.
This also fixes a potential issue where the request timeout for the KATO command does not match the value in the connect command, which might be causing spurious connection drops from the target.Signed-off-by Hannes Reinecke <[email protected]>--- drivers/nvme/host/core.c | 9 ++++++--- drivers/nvme/host/fabrics.c | 2 +- drivers/nvme/host/nvme.h | 9 ++++++++- 3 files changed, 15 insertions(+), 5 deletions(-)drivers/nvme/host/core.c +6
-3@@ -1223,7 +1223,8 @@ static void nvme_keep_alive_end_io(struct request *rq, blk_status_t status) startka = true;spin_unlock_irqrestore (&ctrl->lock, flags);if (startka) -queue_delayed_work (nvme_wq, &ctrl->ka_work, ctrl->kato *HZ ); +queue_delayed_work (nvme_wq, &ctrl->ka_work, +NVME_KATO_DELAY (ctrl->kato) *HZ ); }static int nvme_keep_alive (struct nvme_ctrl *ctrl)@@ -1258,7 +1259,8 @@ static void nvme_keep_alive_work(struct work_struct *work)dev_dbg (ctrl->device,"reschedule traffic based keep-alive timer\n" ); ctrl->comp_seen = false; -queue_delayed_work (nvme_wq, &ctrl->ka_work, ctrl->kato *HZ ); +queue_delayed_work (nvme_wq, &ctrl->ka_work, +NVME_KATO_DELAY (ctrl->kato) *HZ );return ; }@@ -1275,7 +1277,8 @@ static void nvme_start_keep_alive(struct nvme_ctrl *ctrl)if (unlikely (ctrl->kato ==0 ))return ; -queue_delayed_work (nvme_wq, &ctrl->ka_work, ctrl->kato *HZ ); +queue_delayed_work (nvme_wq, &ctrl->ka_work, +NVME_KATO_DELAY (ctrl->kato) *HZ ); }void nvme_stop_keep_alive (struct nvme_ctrl *ctrl)drivers/nvme/host/fabrics.c +1
-1@@ -382,7 +382,7 @@ int nvmf_connect_admin_queue(struct nvme_ctrl *ctrl)* and add a grace period for controller kato enforcement */ cmd.connect.kato = ctrl->kato ? -cpu_to_le32 ((ctrl->kato +NVME_KATO_GRACE ) *1000 ) :0 ; +cpu_to_le32 (ctrl->kato *1000 ) :0 ;if (ctrl->opts->disable_sqflow) cmd.connect.cattr |=NVME_CONNECT_DISABLE_SQFLOW ;drivers/nvme/host/nvme.h +0
-0@@ -27,7 +27,14 @@ extern unsigned int admin_timeout;#define NVME_ADMIN_TIMEOUT (admin_timeout * HZ)
drivers/nvme/host/nvme.h extern unsigned int admin_timeout; · line 37
Quoted 10 lines
drivers/nvme/host/nvme.h +8
-1#define NVME_DEFAULT_KATO 5 -#define NVME_KATO_GRACE 10 + +/* +* The recommended frequency for KATO commands +* according to NVMe 1.4 section 7.12.1: +* The host should send Keep Alive commands at half of the +* Keep Alive Timeout accounting for transport roundtrip times [..]. +*/ +#define NVME_KATO_DELAY (k) ((k) >> 1)
what will happen in case k == 1 ?
drivers/nvme/host/nvme.h extern unsigned int admin_timeout; · line 40
Quoted 3 lines
drivers/nvme/host/nvme.h +0
-0#ifdef CONFIG_ARCH_NO_SG_CHAIN #define NVME_INLINE_SG_CNT 0
_______________________________________________
Linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme
Re: [PATCH 2/3] nvme: sanitize KATO setting
Quoted 73 lines · Max Gurtovoy
Quoted 71 lines · Hannes Reinecke
According to the NVMe base spec the KATO commands should be sent at half of the KATO interval, to properly account for round-trip times.
As we now will only ever send one KATO command per connection we can easily use the recommended values.
This also fixes a potential issue where the request timeout for the KATO command does not match the value in the connect command, which might be causing spurious connection drops from the target.Signed-off-by Hannes Reinecke <[email protected]>---
drivers/nvme/host/core.c | 9 ++++++---
drivers/nvme/host/fabrics.c| 2 +-
drivers/nvme/host/nvme.h | 9 ++++++++-3 files changed, 15 insertions(+), 5 deletions(-)drivers/nvme/host/core.c +4
-2@@ -1223,7 +1223,8 @@ static void nvme_keep_alive_end_io(struct request *rq, blk_status_t status) startka = true; spin_unlock_irqrestore(&ctrl->lock, flags); if (startka) - queue_delayed_work(nvme_wq, &ctrl->ka_work, ctrl->kato * HZ); + queue_delayed_work(nvme_wq, &ctrl->ka_work, + NVME_KATO_DELAY(ctrl->kato) * HZ); } static int nvme_keep_alive(struct nvme_ctrl *ctrl)@@ -1258,7 +1259,8 @@ static void nvme_keep_alive_work(struct work_struct *work) dev_dbg(ctrl->device, "reschedule traffic based keep-alive timer\n"); ctrl->comp_seen = false; - queue_delayed_work(nvme_wq, &ctrl->ka_work, ctrl->kato * HZ); + queue_delayed_work(nvme_wq, &ctrl->ka_work, + NVME_KATO_DELAY(ctrl->kato) * HZ); return; }@@ -1275,7 +1277,8 @@ static void
nvme_start_keep_alive(struct nvme_ctrl *ctrl)
if (unlikely(ctrl->kato == 0))
return;
-queue_delayed_work(nvme_wq, &ctrl->ka_work, ctrl->kato * HZ); + queue_delayed_work(nvme_wq,&ctrl->ka_work,
+NVME_KATO_DELAY(ctrl->kato)* HZ);
}
voidnvme_stop_keep_alive(struct nvme_ctrl *ctrl)drivers/nvme/host/fabrics.c +0
-0@@ -382,7 +382,7 @@ int nvmf_connect_admin_queue(struct nvme_ctrl *ctrl)* and add a grace period for controller kato enforcement */ cmd.connect.kato =
ctrl->kato?
-cpu_to_le32((ctrl->kato + NVME_KATO_GRACE) * 1000): 0; +cpu_to_le32(ctrl->kato * 1000): 0;
if (ctrl->opts->disable_sqflow) cmd.connect.cattr |=NVME_CONNECT_DISABLE_SQFLOW;drivers/nvme/host/nvme.h +2
-1@@ -27,7 +27,14 @@ extern unsigned int admin_timeout; #define NVME_ADMIN_TIMEOUT (admin_timeout * HZ) #define NVME_DEFAULT_KATO 5 -#define NVME_KATO_GRACE 10 + +/*
Quoted 10 lines
drivers/nvme/host/nvme.h extern unsigned int admin_timeout; · line 37
Quoted 6 lines
drivers/nvme/host/nvme.h +6
-0+* The recommended frequency for KATO commands +* according to NVMe 1.4 section 7.12.1: +* The host should send Keep Alive commands at half of the +* Keep Alive Timeout accounting for transport roundtrip times [..]. +*/ +#define NVME_KATO_DELAY (k) ((k) >> 1)what will happen in case k == 1 ?
Ho-hum. Good point.
Will fix it up.
Cheers,
Hannes
Signature · 9 lines
Dr. Hannes Reinecke Kernel Storage Architect [email protected] +49 911 74053 688 SUSE Software Solutions Germany GmbH, Maxfeldstr. 5, 90409 Nürnberg HRB 36809 (AG Nürnberg), GF: Felix Imendörffer _______________________________________________ Linux-nvme mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-nvme
PATCH v2 3/3 nvme: add 'kato' sysfs attribute
In reply to Hannes ReineckeAdd a 'kato' controller sysfs attribute to display the current keep-alive timeout value (if any). This allows userspace to identify persistent discovery controllers, as these will have a non-zero KATO value.
--- drivers/nvme/host/core.c | 2 ++ 1 file changed, 2 insertions(+)
drivers/nvme/host/core.c +2 -0
@@ -3554,6 +3554,7 @@ nvme_show_int_function(cntlid);nvme_show_int_function (numa_node);nvme_show_int_function (queue_count);nvme_show_int_function (sqsize); +nvme_show_int_function (kato); static ssize_t nvme_sysfs_delete(struct device *dev, struct device_attribute *attr, const char *buf,
@@ -3720,6 +3721,7 @@ static struct attribute *nvme_dev_attrs[] = { &dev_attr_hostid.attr, &dev_attr_ctrl_loss_tmo.attr, &dev_attr_reconnect_delay.attr, + &dev_attr_kato.attr, NULL };
Signature · 7 lines
2.29.2 _______________________________________________ Linux-nvme mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-nvme
Re: [PATCH 3/3] nvme: add 'kato' sysfs attribute
_______________________________________________
Linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme
Re: [PATCH 3/3] nvme: add 'kato' sysfs attribute
In reply to Hannes ReineckeQuoted 4 lines · Hannes Reinecke
Add a 'kato' controller sysfs attribute to display the current keep-alive timeout value (if any). This allows userspace to identify persistent discovery controllers, as these will have a non-zero KATO value.
I'm not sure it will identify the persistent discovery controller since you might have non-zero KATO for regular controllers as well.
But for identification you can set special MN or SN for these controllers.
The patch is useful for displaying the keep-alive timeout value.
_______________________________________________
Linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme